Identification of genes to predict optimal T cell receptor function in single cells

Antigen-specific T cells can be effective for the treatment of viral infections and malignancies; however, the in vitro expansion of tumor antigen-specific T-cells is often difficult due to the low frequency of circulating antigen-specific T-cells that are rendered anergic in the immunosuppressive tumor microenvironment. Hence the frequency of tumor antigen-specific T-cells within an infusion product may be low. This problem has been overcome by using recombinant tumor antigen-specific T cell receptors (TCRs) identified from patients. However, the identification of TCRs with optimal binding affinities among potentially thousands of TCRs has not been described. Therefore, to selectively rank TCRs, we identified a gene signature transiently expressed during viral antigen specific T cell activation that we associate with the mechanism of control governing an optimal T cell response. Using this gene signature and a scoring metric against a reference dataset we could predict the specificity of low frequency TCRs and their functional avidity with greater accuracy than predictions made using the canonical marker IFN-γ. We conclude that single cell RNA sequencing ofthe use of single gene expression following single cell RNA sequencing vastly underestimates TCR repertoire specificity and suggest using a multidimensional approach. Using our method, it is possible to rank single cells against any reference and gene set chosen to define optimal response.